DEM Generation Of Moon Based On Line Array CCD Image

With the work of the Moon and the study of deep space exploration,human curiosity about its surface structure and the internal mechanisms is growing.The main research object of this paper is the more accurate data of Chang'e-2 and Chang'e-3.At the same time,the theory and technology of Chang'e data processing,image matching and lunar DEM reconstruction are systematically studied.For Chang'e-2 data,the resolution of the acquired linear array CCD image data is 7 meters,and its accuracy is obviously higher than that of Chang'e-1.Therefore,this paper has done a lot of research on Chang'e-2 image data.Firstly,the DEM reconstruction of the lunar surface is established by Chang'e-2 data.In this research process,the features are extracted by using the improved Forstner operator method,and then the feature points are roughly matched by the pyramid matching method.The results of rough matching are processed by least squares algorithm,and the results of fine matching are obtained.The ultimate goal is to obtain more accurate corresponding image point coordinates,so as to lay the foundation for the selection of the control points.Because Chang'e-3 has the advantage of higher accuracy,but at the same time lacks elevation value,it can not achieve high-precision lunar DEM reconstruction.In view of the above problems,this paper presents a method for estimating the elevation of Chang'e-3.First of all,the use of Chang'e-2 data samples are used for the BP neural network model training.Secondly,the Chang'e-3 data were estimated by the elevation value of mature model.On this basis,the DEM reconstruction of Chang'e-3 can be realized.The innovations of this paper are as follows:1.In the process of using Chang'e-2 lunar data set up in DEM,in order to obtain the 3D coordinates of the image coordinate system,this paper takes after matching the corresponding image points and laser altimeter data into the control point.Then in the lunar coordinate system,the data is coordinate transformed to form three-dimensional coordinate points(x,y,z).According to the matching method,the features are extracted by using the improved Forstner operator method,and then the feature points are roughly matched by the pyramid matching method.The results of rough matching are processed by least squares algorithm,and the results of fine matching are obtained.During the experiment,based on Forstner,Moravec,Harris operator image matching methods are compared and analyzed in detail.It is concluded that the error rate of the matching method based on Forstner operator is reduced to 10.38% in the lunar image matching.2.In view of the lack of elevation value in Chang'e-3 data,this paper has done a lot of work on Chang'e-3 data in order to obtain higher and higher DEM on the lunar surface.The BP neural network model was trained by using the image data of the same location of Chang'e-2 data,and its model was used to estimate the elevation value of the Chang'e-3 image data.The obtained result was compared with the elevation value of the Chang'e-2.Finally,relative error value is obtained.Experiments show that the improved method proposed in this paper has achieved good results in terms of time and results.3.In the above method,the elevation of the Chang'e-3 estimated by the BP neural network is combined with the same-named image point to form a lunar control point.The inverse distance weighted interpolation method is used to interpolate the lunar surface points,and finally the lunar surface DEM reconstruction is realized.